DESCRIPTION:(Adapted from the abstract) This is a proposal to develop and test a Compton scatter correction suitable for application to 3-D (septa out) PET imaging.Septa removal increases scanner sensitivity by a factor of 4.8, but also increases scatter events about a factor of three. The increased sensitivity can result in benefits of dose reduction, faster imaging, or noise reduction, but first it is necessary to have an accurate scatter correction. The goal of this research is to develop a scatter correction algorithm for 3-D PET based on physical principles using both transmission and emission images, and the theory of Compton scattering to calculate the number of single scattered photons in a projection.This accounts for about 73% of the detected scattered coincidences. During the previous funding period model- based corrections were formulated which were accurate to within 12% and could be computed on a current workstation in under 6 minutes, which is only 8% of the time required to reconstruct the whole 3-D set. The investigators propose to further develop the algorithm to include detector efficiency normalization, convolutional correction for multiple scatters, and correction for scatter originating from activity outside the field of view as well as a thin-septa model. The algorithm will be evaluated using the SIMSET Monte Carlo code, phantom imaging, and human studies. Uniformity, contrast recovery and quantitative accuracy of regional estimates will serve as validation measures. The tested algorithm and its documentation will be made available to other investigators.